BACKGROUND OF THE INVENTION For the last few years, cable modem systems based on data-over-cable service specifications have been accepted as a"last mile"high-speed data solution for the consumers.

A two-way Hybrid Fiber-Coax (HFC) cable network is an infrastructure capable of supporting multiple overlaying services, such as analog or digital video service, high- speed data, and telephony service. These services use different band of the available spectrum in the downstream and upstream directions, and each service has its own operations and provisioning infrastructure. At customer premises, a full- service subscription requiring multiple boxes of customer premises equipment (CPE) such as a set top box, a telephone network interface unit, and a cable modem.

These overlaying services result in a costly infrastructure for provisioning and management, as well as higher cost CPEs and inefficient use of upstream spectrum.

Convergent Network

It is therefore highly desirable to have a converged network, capable of delivering voice, video and data in a unified communications infrastructure.

A conventional data-over-cable media-access-control (MAC) protocol is based on sharing an upstream and a downstream channel. Each cable modem or service is statically assigned to the upstream channel. Switching cable modems for load balancing or service balancing among multiple channels is complex and slow.

Moreover, cable modems have severe limitations when it comes to support digital video services. Conventional digital video (broadcast or video on demand) requires more stringent bit-error-rate and quality of service (QOS) than data services. High bit- rate of approximately 20 Mbps per HDTV movie channel is required. Therefore, providing multi-program high-definition digital video services in the same downstream channel used for conventional cable modem data services is inadequate.

Upstream Limitations The upstream bandwidth of the HFC network is limited by two factors: first, the amount of available spectrum in the upstream in a conventional"sub-split"HFC cable plant is between 5 to 42 Mhz in North America. Because of ingress interference, a good portion of the spectrum is not suitable for wide-band (e. g.

3.2Mhz or 6.4 Mhz per channel) and higher-order modulations (e. g. 16,32, or 64 QAM) to achieve high capacity for the upstream channel in use. If a 6.4 Mhz channel is used, only 6. 4/ (42-5) = 17% of the upstream spectrum is used. The other 83% of the spectrum (in particular for frequencies below 10 MHz) is often unused. A conventional data-over-cable MAC is quite limited in the ability to fully utilize the upstream spectrum to maximize the capacity, the efficiency to provide the QoS required by all the services.

Specifically, since each upstream channel must support the packets generated by different services with different QoS requirements, it is very difficult to achieve high

channel utilization under dynamically changing traffic conditions. In particular, the overhead of the MAC management packets such as bandwidth request and initial calibration can be significant andwill complicate the scheduling efficiency of the cable modem termination system (CMTS).

Conventional data-over-cable MAC protocol relies on some form of polling to achieve QoS goal of meeting bandwidth, latency and jitter requirements. For a polling interval of 2ms, each upstream channel requires about 270 Kbps of downstream bandwidth for the MAC operation. This represents a significant amount of bandwidth from the downstream channel. Therefore, scalability of using multiple upstream channels in conventional data-over-cable is quite limited.

Broadcast Quality Digital Video Although the HFC network has sufficient bandwidth to support delivery of a full spectrum of services including data, telephony and video, these services currently are separate infrastructures provisioned by a service provider. As a result, these are sub-optimal usage of the HFC spectrum and costly duplication of equipment at the head end and at customer premises. Voice-over-Internet Protocol enables convergence of voice and data. However, video service remains a separate infrastructure.

Therefore, there is an unmet need for a unified communications system that can provide the full need of providing broadband Internet access, IP telephony, broadcast quality digital video over the same HFC system.

Therefore, there is an unmet need for a MAC that can be used to implement a full service cable modem system to fulfill the full potential of a HFC network for delivery voice video and data cost-effectively to the home and business.

It will be realized, after the detailed description of the invention, how to overcome the limitations of conventional cable modem systems.

The full-service MAC described herein fully utilizes the upstream and downstream spectrum of a conventional HFC cable plant, enabling service providers economically to deploya full spectrum of services with voice, video and data, without a forklift upgrade to the HFC cable plant The unified full service communications system described herein will drastically reduce cost of providing three separate provisioning systems for video, data and voice, at the head end and at the same time reduce the number of on-premises equipment from three to one.

It is an object of the present invention to overcome the disadvantages of the prior art.

BRIEF SUMMARY OF THE INVENTION This and other objects are achieved by the present invention. In accordance with the present invention a full-service cable modem (fsCM) system capable of delivering video, data and voice over a two-way hybrid fiber-coaxial cable network is described.

A high-capacity, high-efficiency multi-channel full-service MAC, capable of supporting multiple upstream and downstream channels, enables the fsCM system 100 to deliver a full spectrum of services presently requiring multiple delivery systems. The video's can be delivered by a combination of highquality broadcast MPEG-2 audio/visual streams and Internet Protocol (IP) video streams., Further, multiple channels can be used to multiplex packets of all types, enabled by a true seamless channel change described in this invention, maximizing the statistical multiplexing gain. Packet-by-packet channel switching enables fast recovery from a channel failure, as required by a high-availability fault-tolerance voice-over-IP telephony service in the cable modem system.

The fsCM system 100 consists of, according to the preferred embodiment, illustratively two downstream channel (DCPC and DPC1), two upstream payload channels (UPC1 and UPC2), three upstream control channels (UCC1, UCC2, UCC3)

in the HFC cable plant that connect a fsCMTS in the head-end and a plurality of fsCMs at subscriber sites.

A fsCM uses DCPC for downstream MAC management messages as well as for payloads (MPEG-2 Transport Stream (TS) or IP packets) and DPC1 for downstream payload channel to deliver high quality MPEG-2 video or IP packets.

The present invention further includes downstream MAC management messages MMAP 900 and MDCD 1000 to enable fsCMTS to allocate upstream transmission to any of the multiple upstream channels on a packet-by-packet basis, and allow a multiple-channel MAC domain to be changed quickly to adapt to changing traffics in the network.

The methods and apparatus described herein implement a novel and unique facility that provides for efficient access of a full-service cable modem network capable of simultaneously servicing the communications needs of internet access, telephony, interactive and on-demand digital video to a large number of users over a conventional HFC network.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram illustrating an embodiment a full-service Cable Modem System; FIG. 2 is a block diagram of a full-service cable modem; FIG. 3 is a diagram illustrating the frequency channel plan for an example full-service cable modem system; FIG. 4 is a block diagram illustrating the structure of SYNC message 500; FIG. 5 is a block diagram illustrating the structure of CREQ message 600; FIG. 6 is a block diagram illustrating the structure of CRSP message 700; FIG. 7 is a block diagram illustrating the structure of BREQ message 800; FIG. 8 is a block diagram illustrating the structure of MMAP message 900; FIG. 9 is a block diagram illustrating the structure of MDCD message 1000; FIG. 10 is a flow diagram illustrating a typical fsCM initialization process; and

FIG. 11 is a flow diagram illustrating a typical upstream data transmission process using contention BREQ 800 and MMAP 900.

DETAILED DESCRIPTION OF THE INVENTION Refer to FIG. 1 for a preferred embodiment of a multi-channel fsCM system 100. A fsCMTS 102, typically located at a head end 101, is connected to the fiber-part of a two-way HFC network 104 through an electrical to fiber interface (not shown). A remotely located fsCM 106 is connected to a coax 402 part of the HFC 104. The downstream spectrum (typically 50 to 850 Mhz) is divided into typically 6 Mhz channels in the downstream for NTSC cable systems. The upstream spectrum typically ranges from 5 to 42 Mhz in North America, and the upstream channel bandwidth varies typically from 160KHz to 6.4 Mhz. The architecture and topology of a modern two-way HFC cable plant are known in the art and will not be repeated here.

In this example, also referring to FIG. 3, there are two downstream channels : a downstream control and payload channel DCPC 147 and a downstream payload channel DPC1 137, and five upstream channels : upstream control channels UCC1 174, UCC2 176, UCC3 178 and upstream payload channels UPC1 182 and UPC2 184. The exemplified channel frequencies are illustrated in FIG. 3, in which channel center frequencies for DCPC 147, DPC1 137, UCC1 174, UCC2 176, UCC3 178, UPC1 182 and UPC2 184 correspond to f1, f2, f3, f4, f5, f6, f7 respectively. The center frequencies for DCPC 147 and DPC1 137 are controlled by the corresponding frequency-agile up-converters 146 and 136. The UCC's 174,176 and 178 channel center frequencies and channel bandwidths are controlled by a burst transmitter 194.

The UPC's 182 and 184 center frequencies and channel bandwidths are controlled by another burst transmitter 196. Illustratively, the UCC's use narrower channel bandwidths and robust modulation schemes such as QPSK or BPSK that can be located in the noisier portion of the upstream spectrum. The"cleaner"part of the upstream spectrum are normally used by UPC's so that higher order of modulations such as 16 to 64 QAM can be used reliably for higher throughput for payloads. In an

alternative embodiment, a single upstream frequency-agile programmable burst transmitter can multiplex the transmission of control and payload bursts.

Through an IP network interface 122, a fsCMTS 102 is connected to a video server 108 for MPEG-2 digital video services, to a managed Internet backbone 112 for connection to a Public Switched Telephone Network PSTN 113, or other voice- over- ! ? networks for telephony services, to an Internet backbone 114 for high-speed data services, and to an Intranet IP network 116 for access to provisioning and network management servers as part of the fsCMTS system operation. The IP network interface is also connected to the video server 108 for providing IP connectivity for video-related network management and illustratively, for upstream traffic generated by a set-top box 530.

Digital video traffics, generated by the video server 108, and packetized into MPEG- 2 transport streams TS 150,152 are multiplexed with fsCMTS MAC messages 160 including SYNC 500, CREQ 600, CRSP 700, BREQ 800, MMAP 900, MDCD 1000 and IP payload packets 154 in downstream transmitters 132,142, which are outputted to downstream modulators 134,144. The intermediate frequency outputs of the modulators 134,144 are up-converted to the desired center frequencies by the up-converters 136 and 146. The radio frequency RF outputs of the up converters 136 and 146 are then transmitted through the HFC plant 104 into downstream receivers 470,420 of the fsCM's 106 via the coaxial 402 portion of the HFC 104.

The downstream modulators 134,144 typically are specified to comply with ITU J83 Annex A, B, or C depending on nationality. Other modulation and forward error correction (FEC) formats are possible.

IP packets 154 are encapsulated in MPEG2-TS using a unique packet identifier PID (1 FFE hexadecimal for data-over-cable) before transmitting downstream.

The time base in the fsCMTS 102 and in the remote fsCMs 106 are synchronized by periodically sending a captured time-stamp value of a time-stamp counter 130 driven by a time-stamp frequency source 128. The time-stamp value is encapsulated in the MAC management message (SYNC 500), which is in turn encapsulated into a

MPEG2-TS and merged with the other TS before delivering to the downstream modulator 134. The method of synchronization using time-stamped message is known in the art. The SYNC 500 is transmitted in all downstream channels so as to enable seamless switching of downstream channels.

The DCPC 147 carries MAC management messages including the MMAP 900 and the MDCD 1000, which are essential for the multi-channel MAC operation, and their significance will be understood when they are described in detail below.

A full-service MAC (fsMAC) has two parts: a fsMAC-CM 192 and a fsMAC-CMTS 124, which are located in the fsCM 106 and the fsCMTS 102 respectively. The fsMAC's role is to co-ordinate the dispatch of downstream IP packets and fsMAC management messages; another role is to co-ordinate the efficient and orderly transmission of upstream bursts using the two upstream burst transmitters 194 and 196.

One of the transmitters 194 is used for transmitting fsMAC management packets such as calibration and bandwidth requests. The other transmitter 196 is for transmitting payload of IP packets 199 received from the CPE interface 197.

More specifically, the transmitter 194 is used to transmit bursts tothe UCC1 174, the UCC2 176 or the UCC3 178 using burst profiles communicated to the fsMAC-CM 106 by the fsMAC-CMTS 124 by sending down MDCD 1000. Similarly, the transmitter 196 is used to transmit bursts to the UPC1 182 or the UPC2 184 using other burst profiles. The fsCM 106 learns the characteristics of burst profiles by listening to the MDCD message 1000 and uses the burst profile and time to transmit by decoding the MMAP message 900.

At the fsCMTS 102, corresponding to these transmitters in the fsCM 106, there are i matching frequency-agile programmable burst receivers consisting an UCC burst receiver 172 and an UPC burst receiver 180, that will receive, demodulate and recover the packets received. These packets (including collision detection information, if any) will be inputted to the fsMAC-CMTS 124.

Full-Service Cable Modem FIG. 2 is a block diagram illustrating an embodiment of the fsCM 106. The RF signal enters the fsCM 106 at the coax 402. The RF is divided into two paths by RF splitter 404. RF paths 406,405 from the splitter 404 are connected to diplex filters 410,460 respectively. The diplex filter 410 passes high frequency downstream RF signal 412 to DPC1 downstream receiver 420, whose output is a MPEG-2 transport stream TS 422 into a packet identifier (PID) de-multiplexing unit 424. The de-multiplexing unit 424 separates a data-over-cable TS 426 from a conventional audio/video/data TS 423 by examining the PID value. The data-over-cable TS 426 is identified by a value of 1 FFE (hexadecimal). The audio/video/data TS 423 associated with a program (e. g. movie) is directed to a conventional MPEG-2 decoder 428 for generating audio/visual signals. Outputs from the decoder 428 can be of a digital television DTV 430, or a standard analog signal 434 (composite video or NTSC modulated RF) for connection to conventional television receivers or video monitors.

Alternatively, the TS 423 can interface to a digital set-top box using IEEE 1394 (not shown), or other high-speed connections. Another alternative is to send a MPEG-2 audio/video/data TS 476 to FSMAC-CM 192, where the TS is encapsulated in IP (MPEG-2 over IP) and forwarded to a home network 508 via the CPE interface 197.

A digital set-top box 503 attached to the home network 508 can decode the MPEG-2 TS.

The other RF path 405 passes through the diplex filter 460. The downstream RF signal 462 is tuned to DCPC 147 and processed by a second downstream receiver 470, whose output is another MPEG-2 transport stream TS 472, which is inputted to the PID demux unit 424, which in turn separates the data-over-cable TS 476 from another audio/video/data TS 473.

The data-over-cable TS 426 is processed in a downstream processing unit 502 to recover data-over-cable packets, consisting of MAC messages and IP payload packets, before entering the fsMAC-CM 192. MAC messages are processed by

fsMAC-CM 192. IP payload packets are forwarded to CPE devices attached to the home network 508, subjected to filtering rules by the CPE interface 197, which is illustratively, an Ethernet network interface. Specifically. IP packets are subjected to filtering rules in the packet forwarding engine within CPE interface 197 using bridging or routing rules. The IP packets are forwarded to CPE devices such as a personal computer 514, an Internet Appliance 512, a Multimedia Terminal Adaptor 516 for voice-over-IP telephony 518, FAX 522, video conferencing 520 and other media streaming services using a home networking infrastructure 508 (e. g. 10/100 Base-T Ethernet, USB, HPNA, Wireless LAN, HomePlug etc. ) Upstream IP packets from the CPE devices 512,514, 516,530 are subjected to filtering by the packet forwarder within the CPE interface 197, and then are queued at an upstream processing unit 586. There are two upstream burst transmitters in this embodiment: an Upstream Control Channel (UCC) burst transmitter 194 and an Upstream Payload Channel (UPC) burst transmitter 196. Each of the two transmitters consists of FEC encoder, modulator, frequency agile digital up converter, RF front-end, etc. to enable upstream burst transmissions in any channel in the upstream spectrum, according to the stored burst profiles sent from the fsCMTS 102.

Upstream MAC management burst packets 498 are sent to the UCC channel transmitter 194, which outputted as a RF burst signal 490 to the diplex filter 460.

Payload IP packets 488 emerge from upstream processing unit 506, accordingly processed by the UPC burst transmitter 196, whose output burst RF signal 480 is coupled to the diplex filter 410 and emerges as the RF signal 406, which is coupled to the HFC coax 402 by the splitter 404, traveling upstream to the head end where the fsCMTS 102 is located.

Now the signal flow of the fsCM system 100 between the fsCMTS 102 and the fsCM 106 has been described. The following further description will show how the fsMAC- CMTS 124 and the fsMAC-CM 192 will coordinate the multiple access transmission of upstream bursts. Essential MAC management messages SYNC 500, MDCD

1000, MMAP 900, CREQ 600, CRSP 700, BREQ 800 are described first and then the fsMAC protocol details will follow.

Full-Service MAC Management Messages 'SYNC Message FIG. 4 is a block diagram of a SYNC MAC message structure 500. The SYNC MAC message structure 500 includes a MAC management header 582, a time stamp snapshot 584 that captures the value of the sampled value of the time stamp counter 130, a fsMAC domain identifier 586, and a downstream channel identifier 588.

A description of the fields of SYNC message 500 is shown in Table 1. However, fewer or additional fields could also be used in the SYNC message 500.

TABLE 1. SYNC MESSAGE 500 Field Parameter Description of Field Parameter fsMAC Message Header 582 This field allows fsCM-MAC 192 to uniquely identify and process the SYNC management message 500.

Time stamp snapshot 584 This field contains the sampled value of time stamp counter 130. fsMAC domain identifier 586 This field uniquely identifies the fsMAC domain as defined by MMAP message 900.

Downstream Channel identifier This field uniquely identifies the downstream channel to 588 which fsMAC messages are transmitted.

'CREQ Message FIG. 5 is a block diagram of a calibration request (CREQ) MAC message structure 600. The CREQ MAC message structure 600 includes a MAC management header 602, a fsCM service identifier 604, a fsMAC domain identifier 606, a downstream

channel identifier 608, a fsCM Ethernet MAC address 610, a fsCM type 612, and a pre-equalizer training sequence 614.

A description of the fields of the CREQ message 600 is shown in Table 2. However, fewer or additional fields could also be used in the CREQ message 600 in other embodiments.

TABLE 2. CREQ MESSAGE 600 Field Parameter Description of Field Parameter fsMAC Message This field allows the fsCM-MAC 192 to uniquely identify and Header 602 process the CREQ message 600. fsCM service identifier This field uniquely identifies the service flow associated with the (SID) 604 fsCM 106 within the fsMAC domain identified by the fsMAC domain ID 606 fsMAC domain This field uniquely identifies the fsMAC domain as defined by identifier (MAC ID) 606 MMAP message 900.

DCPC channel identifier This field uniquely identifies the downstream control and payload 608 channel (DCPC) into which fsMAC messages are transmitted Ethernet MAC address This field contains the 48-bit Ethernet MAC address associated 610 with the fsCM 106 fsCM type 612 This field contains information about the type and version of the fsCM 106 Pre-equalizer training This field contains pre-equalizer training sequence (s) for the sequence 614 fsCM transmitters 194,196.

CRSP Message FIG. 6 is a block diagram of a calibration response MAC message structure 700. The CRSP MAC message structure 700 includes a MAC management header 702, a fsCM service identifier 704, a fsMAC domain identifier 706, an upstream channel identifier 708, a timing adjustment 710, a frequency adjustment 712, a transmit

power adjustment 714, transmitter pre-equalizer tap coefficients 716, and a re- assigned fsMAC domain identifier 718.

A description of the fields of the CRSP message 700 is shown in Table 3. However, fewer or additional fields could also be used in the CRSP message 700 in other embodiments.

TABLE 3. CRSP MESSAGE 700 Field Parameter Description of Field Parameter fsMAC Message Header 702 This field allows the fsCM-MAC 192 to uniquely identify and process the CRSP message 700. fsCM service identifier (SID) 704 This field uniquely identifies the service flow associated with the fsCM 106 within the fsMAC domain identified by the fsMAC domain ID 706 fsMAC domain idenifier (MAC ID) This field uniquely identifies the fsMAC domain as 706 defined by a MMAP message 900.

Upstream channel identifier 708 This field identifies the upstream channel CRSP 700 is responding to.

Timing adjustment 710 This field contains information for the fsCM 106 to adjust its local clock to synchronize with that of fsCMTS Frequency adjustment 712 This field contains information for the fsCM 106 to adjust its upstream transmitter center frequency to within the receiving frequency range of the fsCMTS receiver.

Transmit power adjustment 714 This field contains information for the fsCM 106 to adjust its transmitter power amplifier gain to the correct level.

Transmit pre-equalizer tap This field contains information for the fsCM 106 to coefficients 716 adjust its transmitter pre-equalizer to this new parameters.

Reassigned fsMAC domain This field contains information (if present) about a new identifier 718 fsMAC domain identifier, which the fsCM 106 will

associate with after receiving this message.

BREQ Message FIG. 7 is a block diagram of a bandwidth request (BREQ 800) MAC message structure 800, which includes a fsMAC message header 802, a fsCM service identifier 804, a fsMAC domain identifier 806, a framing header type 808, and an amount requested 810.

A description of the fields of the BREQ message 800 is shown in Table 4. However, fewer or additional fields could also be used.

TABLE 4. BREQ MESSAGE 800 Field Parameter Description of Field Parameter fsMAC Message Header 802 This field allows the fsCM-MAC 192 to uniquely identify and process the BREQ message 800. fsCM service identifier (SID) 804 This field uniquely identifies the service flow associated with the fsCM 106 within the fsMAC domain identified by the fsMAC domain ID 806 fsMAC domain idenifier (MAC ID) This field uniquely identifies the fsMAC domain as 806 defined by MMAP message 900.

Framing header type 806 This field contains the header type information for the fsCMTS to take into consideration of the MAC frame header overhead when allocating bandwdith for the requesting fsCM.

Amount requested 810 This field contains amount of payload bandwidth (excluding MAC header overhead) requested by the fsCM, e. g. number of bytes or number of time slots such as mini-slots.

MMAP message

FIG. 8 is a block diagram of a multi-channel bandwidth allocation MAC message (MMAP) structure 900, which includes a fsMAC management message header 902, a fsMAC domain identifier 904, a list of broadcast grants, a list of unicast grants, and a list of pending grants 910.

A description of the fields of the MMAP message 900 is shown in Table 5. However, fewer or additional fields could also be used.

TABLE 5. MMAP MESSAGE 900 Field Parameter Description of Field Parameter fsMAC Message Header 902 This field allows the fsCM-MAC 192 to uniquely identify and process the MMAP message 900. fsMAC domain identifier 904 This field uniquely identifies the fsMAC domain Broadcast grants 906 This field contains the bandwidth grants for the contention area that bandwidth requests are transmitted from any fsCM in the fsMAC domain. Table 6 gives an example of the broadcast grants Unicast grants 908 This field contains the bandwidth grants address to an individual fsCM. Table 7 gives and example of unicast grants.

Pending grants 910 This field contains a list of pending grants for those BREQ's that are successfully received by the fsCMTS, but the grants are deferred to a later MMAP 900. Table 8 gives an example of pending grants TABLE 6. Broadcast grants 906 example Broadcast Grants Description of Field Parameter Number of broadcast grants =2 in this example Service ID (Start of 15'broadcast grant). This field contains the SID of the broadcast address for all fsCM's.

Grant type Bandwidth request BREQ 800 Upstream channel ID This field contains the channel ID to which the broadcast grant is allocated Burst profile ID This field identifies the burst profile of BREQ 800 Back-off start and End values This field contains the back-off window of the chosen contention resolution algorithm Length of payload data in bytes BREQ 800 burst payload data length in bytes Number of bursts Number of BREQ 800 bursts for this grant Transmission start time Start transmission time of the first BREQ 800 burst Service ID (Start of 25t broadcast grant). This field contains the SID of a broadcast address for a group of fsCM's.

Grant type Calibration request CREQ 600 Upstream channel ID This field contains the channel ID to which the broadcast grant is allocated Burst profile ID This field identifies the burst profile of the CREQ 600 Back-off start and End values This field contains the back-off window of the chosen contention resolution algorithm in this example Length of payload data in bytes CREQ 600 burst payload data length in bytes Number of bursts Number of CREQ 600 bursts for this grant Transmission start time Start transmission time of the first CREQ 600 burst TABLE 7. Unicast grants 906 example Unicast Grants Description of Field Parameter Number of Unicast grants 3 in this example SID-1 (Start of 1st unicast grant). This field contains SID of fsCM-1.

Grant type Variable length payload packet Upstream channel ID This field contains the channel ID to which the unicast grant is allocated Burst profile ID This field identifies the burst profile for packet Burst framing header type This field contains framing header type to enable fsCMTS to calculate the overhead needed for the burst Length of payload data in bytes Burst payload data length in bytes Transmission start time Start transmission time of the first BREQ 800 burst SID-2 (Start of 25t unicast grant). This field contains SID of fsCM-2.

Grant type Constant bit rate (CBR) Upstream channel ID This field contains the channel ID to which the unicast grant is allocated Burst profile ID This field identifies the burst profile for this burst Burst framing header type This field contains framing header type to enable fsCMTS to calculate the overhead needed for the burst Length of payload data in bytes Burst payload data length in bytes Grant interval This field contains the time interval between two adjacent grants Transmission start time Start transmission time of the burst SID-3 (Start of 3"unicast grant). This field contains SID of fsCM-3.

Grant type Dedicated channel Upstream channel ID This field contains the channel ID to which the unicast grant is allocated Length of payload data in bytes Burst payload data length in bytes Grant duration This field contains the time for which the dedicated channel can be used Transmission start time Start transmission time of the first burst

TABLE 8. Pending grants 910 example Pending Grants Description of Field Parameter Number of broadcast grants =2 in this example SID-a This field contains the SID of the pending grant for fsCM-a.

SID-b This field contains the SID of the pending grant for fsCM-b.

0 MDCD Message FIG. 9 is a block diagram of a fsMAC domain channel descriptor (MDCD) MAC message structure 1000, which includes a fsMAC message header 1002, a fsMAC domain identifier 1004, an accept new fsCM registration flag 1006, a number of downstream channels 1008, a number of upstream channels 1010, a downstream channel change count 1012, an upstream channel change count 1014, a list of downstream channel identifiers and Type-Length-Values (TLV's) 1026, a list of upstream channel identifiers and TLV's 1028, and a list of upstream burst profile identifiers and TLV's 1030.

A description of the fields of MDCD message 1000 is shown in Table 9. However, fewer or additional fields could also be used.

TABLE 9. MDCD MESSAGE 1000 Field Parameter Description of Field Parameter fsMAC Message Header 1002 This field allows fsCM-MAC 192 to uniquely identify and process the MDCD message 1000. fsMAC domain identifier 1004 This field uniquely identifies the fsMAC domain as defined by the MMAP message 900.

Accept-new-fsCM-registration flag This field contains a flag bit which when set, indicating 1006 the fsMAC domain is accepting the new fsCM 106 registration.

Number of downstream channels This field contains N number of downstream channels 1008 in the fsMAC domain.

Number of upstream channels 1010 This field contains M number of upstream channels in the fsMAC domain.

Downstream channel change count This field contains a count of changes in downstream 1012 channel configuration. If this field is different than the count in the previous MDCD message 1000, the fsCM's 106 in the fsMAC domain must update its downstream channel configuration to the current MDCD message 1000.

Upstream channel change count This field contains a count of changes in upstream 1014 channel configuration. If this field is different than the count in the previous MDCD message 1000, the fsCM's 106 in the fsMAC domain must update its upstream channel configuration to the current MDCD message 1000.

List of downstream channel This field contains a list of N downstream channel identifiers and TLV's 1026 identifiers and the associated TLV's defining the channel parameters. Table 10 shows an example of a list of 2 downstream channels.

List of upstream channel identifiers This field contains a list of M upstream channel and TLV's 1028 identifiers and the associated TLV's defining the channel parameters. Table 11 shows an example of a list of 5 upstream channels.

List of upstream burst profile This field contains a list of X upstream burst profile identifiers and TLV's 1030 identifiers and the associated TLV's defining the burst parameters. Table 12 shows an example of a list of 3 burst profiles.

TABLE 10. Downstream channel identifiers and TLV's 1026 example Number of downstream TLV encoding channels = 2 downstream channel parameter Type Length Value Description type (1 byte) (1 byte) (L bytes) Downstream channel identifier 1 1 01 01 (Channel ID) Downstream channel type 2 1 1 1 (DCPC) Center frequency 3 4 fl Hz Symbol rate 4 1 0 0 (5. 056941 Msymbols/sec) FEC 5 1 1 1 (J83 Annex B) Modulation 6 1 0 64QAM Interleave depth (I, J) 7 2 16,8 Latency=0. 48 ms Downstream channel identifier 1 1 02 02 Downstream channel type 2 1 2 2 (DPC 1) Center frequency 3 4 f2 Hz Symbol rate 4 1 1 1 (5.360537 Msymbols/sec) FEC 5 1 1 1=J83 Annex B Modulation 6 1 1 256QAM Interleave depth (I, J) 7 2 128,1 Latency=2. 8ms TABLE 11. Upstream channel identifiers and TLV's 1028 example Number of upstream channels TLV encoding =5 Upstream channel parameter Type Length Value Description type (1 byte) (1 byte (L bytes Upstream channel identifier 1 1 10 10 Upstream channel type 2 1 0 0 (UCC 1) Center frequency 3 4 f3 Hz Symbol rate 4 1 3 3 (640 Upstream channel identifier 1 1 11 Channel ID =11 Upstream channel type 2 1 1 1 (UCC2) Center frequency 3 4 f4 Hz Symbol rate 4 1 2 2 (320 Ksymbols/sec) Upstream channel identifier 1 1 12 Channel ID =12 Upstream channel type 2 1 2 2 (UCC3) Center frequency 3 4 f5 Hz Symbol rate 4 1 3 3=640 Ksymbols/sec Upstream channel identifier 1 1 13 Channel ID =13 Upstream channel type 2 1 3 3 (UPC1) Center frequency 3 4 f6 Hz Symbol rate 4 1 6 6=5.12 Msymbols/sec Upstream channel identifier 1 1 14 Channel ID =14 Upstream channel type 2 1 4 4 (UPC2) Center frequency 3 4 f7 Hz Symbol rate 4 1 6 6=5.12 Msymbols/sec TABLE 12. Upstream burst profile identifiers and TLV's example Number of upstream burst TLV encoding profiles = 3 upstream burst parameter Type Length Value Description type (1 byte) (1 byte) (L bytes) Burst identifier 1 1 11 Burst profile 1 Modulation 2 1 0 0=QPSK Preamble length 3 2 64 64 bites FEC code word (k) 4 1 78 13 bytes FEC error correction (T) 5 1 6 T=2 bytes Scramble seed 6 2 35 Seed =00110101 Inter-burst guard time 7 1 5 5 symbols burst identifier 1 1 12 Burst profile 2 modulation 2 1 0 0=QPSK Preamble length 3 2 64 64 bites FEC code word (k) 4 1 78 78 bytes FEC error correction (T) 5 1 6 T=6 bytes

Scramble seed 6 2 35 Seed =00110101 Inter-burst guard time 7 1 5 5 symbols burst identifier 1 1 13 Burst profile 3 Modulation 2 1 0 0=64QAM Preamble length 3 2 64 128 bites FEC code word (k) 4 1 78 256 bytes FEC error correction (T) 5 1 6 T=10 bytes Scramble seed 6 2 35 Seed =00110101 Inter-burst guard time 7 1 5 5 symbols Full-Service Cable Modem System Operation For this exemplified embodiment, the fsCMTS sets up the fsCM domain consisting of: 2 downstream channels 1. The DCPC 147 is a broadcast channel for all fsCMs 106 within the fsCM domain and is configured to ITU-T J83 Annex B standard with a 64QAM modulation and at a center frequency of f1 Hz in the downstream spectrum as shown in FIG. 3. This channel is is primarily for data-over-cable MAC management messages, IP traffic and to a less extent, MPEG-2 video delivery.

2. The DPC1 137 is the a broadcast channel for all fsCMs 106 within the fsCM domain, is configured to be ITU-T J83 Annex B standard with a 256QAM modulation and at a center frequency of f2 Hz in the downstream spectrum as shown in FIG. 3. This channel is primarily for broadcast quality MPEG-2 movie delivery, but also carries IP packets.

3 upstream control channels 1. The UCC1 174 for contention bandwidth request for all or a group of fsCMs 106, is configured to operate at 640 Ksymbols/sec with QPSK modulation and at a center frequency of f3 Hz in the upstream spectrum as shown in FIG. 3.

2. The UCC2 176 is for contention calibration and maintenance for all or a group of fsCMs 106, and is configured to operate at 320 Ksymbols/sec

with a QPSK modulation and at a center frequency of f4 Hz in the upstream spectrum as shown in FIG. 3.

3. The UCC3 178 is for Aloha contention, pay-per-view or video-on- demand request bursts for all or a group of fsCMs 106, and is configured to operate at 640 Ksymbols/sec with a QPSK modulation and at a center frequency of f5 Hz in the upstream spectrum as shown in FIG. 3.

2 upstream payload channels: 1. The UPC1 182 is intended primarily for voice-over-IP CBR traffic for all or a group of fsCMs 106, and is configured to operate at 5.12 Msymbols/sec with a16QAM modulation and at a center frequency of f6 Hz in the upstream spectrum as shown in FIG. 3.

2. The UPC2 184 is intended primarily for high-speed data and media streaming traffic for all or a group of fsCMs 106 is configured to operate at 5.12 Msymbols/sec with 16QAM modulation and at center frequency off7 Hz in the upstream spectrum as shown in FIG. 3.

When the fsCMTS 102 is operational, the following MAC management messages are broadcast periodically to all fsCMs 106 to establish a fsCM domain, in the HFC 104 via the DCPC 147: 1. the SYNC 500, typically sent every 150 to 250 ms, 2. the MDCD 1000, typically sent every 1 to 2 seconds, and 3. the MMAP 900, typically sent every 2 to 10 ms.

The SYNC 500 establishes network-wide clock synchronization of the fsCMTS 102 and fsCMs 106 using a conventional time-stamp methodology and is known in the art. The MDCD 1000 establishes the fsMAC domain using the fsMAC domain identifier 1004. The MDCD 1000 also contains the parameters needed by the fsCMs 106 to join the fsMAC domain by setting up the channel and burst profiles. The MMAP 900 contains information about upstream transmission opportunities on a specific channel, using a specific burst profile, duration of the transmission, and at a specific start time to transmit. The MMAP 900 also contains upstream transmission opportunities, typically once every 1 to 2 seconds, for the fsCM 106 that wishes to join the network to transmit the CREQ 600 to adjust its ranging offset, center

frequency, transmitter power level, and transmitter pre-equalizer coefficients as part of the initialization process. Once initialized, the fsCM 106 starts to use the contention-based BREQ 800 to request transmission of payload packets.

FulKService Cable Modem Initialization Referring to FIG. 10, a fsCM initialization flow diagram 1100 is entered at a block 1102 when the fsCM 106 is powered up or reset. In a block 1104, the DCPC receiver 470 at the fsCM 106 is continuously searching for a valid DCPC channel. The DCPC is considered as valid if MPEG-2 TS with a valid data-over-cable PID (e. g. 1 FFE hexadecimal) and once found, a block 1106 is entered to search for a valid MDCD 1000. In the MDCD 1000, the flag 1006, if set, signifies that the DCPC is accepting the new fsCM 106 registrations, and a block 1110 is entered. If the flag 1006 is not set, signifying the MDCD 1000 is not taking in new registrations, the fsCM 106 will exit the block 1106 and enter the block 1104 for searching for another valid DCPC.

In the block 1110, all the parameters in the MDCD 1000 are accepted by the fsCM 106. The fsMAC domain ID 1004 will be used to match the domain identifier 586 in the SYNC 500. If the valid SYNC 500 is received, the fsCM 106 will synchronize its time base with the fsCMTS time base in a block 1114.

The fsCM 106 also initializes the other downstream and upstream channels, and the burst profiles, based on information received in the MDCD 1000 in the block 1114.

In a block 1116, the fsCM 106 also monitors the MMAP 900 for the broadcast calibration grant as shown in Table 6. In this example, the second broadcast grant is for the CREQ 600. In a block 1116, If the CREQ 600 grant is received a block 1118 will be entered, and the fsCM 106 will construct a calibration burst based on the burst profile, and length of payload information as specified in the received broadcast grant 906.

In a block 1120 the CREQ 600 burst will then be transmitted at the specified upstream channel, and at the specified transmission start time (subject to back-off based on the back-off start and end values specified in the grant using exponential back-off algorithm). If the calibration response CRSP 700 is received by the fsCM 106 in a block 1122, the initial calibration is successful and a fine calibration block 1124 is entered. If the CRSP 700 is not received in the block 1122, after a pre-determined time-out, the block 1116 will be entered and the CREQ 600 process will be retried (not shown).

In the block 1124, the fsCMTS will perform fine calibration on each of the upstream channels in the fsCM domain by sending a periodic unicast fine calibration grant to the fsCM 106 for each upstream channel. In the block 1124 the fine-calibration process is complete after receiving fine the CRSP 700 from the fsCMTS 102 and after the fsCM 106 adjusting its upstream channel parameters including ranging offset, frequency, power level, and pre- equalizer coefficients. These parameters will be saved in the fsCM 106 upstream channel profiles and they will be used to configure the channel before burst transmission. After fine calibration, a block 1126 is entered. The fsCM 106 completes the modem registration process and becomes operational in a block 1128.

Transmission Using Bandwidth Request Referring to FIG. 11, which is a flow diagram of transmission using a contention-based bandwidth request 1200. In a block 1204, one or more packets are queued up at the fsCM 106. In a block 1206 the fsMAC-CM 192 chooses to transmit one or more of packets. The number of bytes of payload and header type (e. g. short, long or concatenated) are determined. In a block 1208, the fsCM 106 waits until the MMAP 900 is received with the broadcast grant 906 in the BREQ 800 (example in Table 6). Entering a block 1210, the fsCM 106 uses the back-off start and end values to calculate the initial back- off of burst transmission (any back-off algorithm will work and is well-known in the art). If the back-off algorithm determines the transmission opportunity is

beyond the current grant, the fsCM 106 will defer the transmission to the next MMAP 900; otherwise, referring to the 1 broadcast grant in Table 6,, the fsCM 106 calculates the BREQ 800 burst transmission start time according to: (Transmission start time) + (Burst duration calculated and based on the length of payload and header in bytes and burst profile) x (number of burst deferred calculated by the back- off algorithm).

The BREQ 800 will be transmitted at the calculated time at the channel specified by the upstream channel ID. A block 1212 is entered and the fsCM 106 waits for a unicast grant or a pending grant in the next MMAP 900. The next MMAP 900 is received in a block 1218 and is checked for a unicast grant with a SID corresponding to the one in the original BREQ 800 in a block 1220.

The unicast grant will have the necessary information (burst profile, header type, and burst profile) to assemble a burst in a block 1226 and transmit at the specified upstream channel at the specified transmission start time (subject to back-off) in a block 1228. If in the block 1220, no unicast grant is received for the BREQ 800, the MMAP 900 is checked for existence of a pending grant.

In a block 1224, if there is a pending grant, the block 1208 is entered to wait for the next MMAP 900. If in the block 1224 there is no pending grant in the MMAP 900, the CREQ 600 is considered as lost or collided, and the block 1208 is re-entered to retry the BREQ 800 transmission.

True Seamless Channel Change In a conventional data-over cable system, a conventional cable modem termination system (CMTS) may direct a cable modem (CM) to change its upstream channel for traffic load balancing, noise avoidance, or failed channel backup. The procedure for performing a channel change is as follows. When the CMTS determines to move a CM from the currently assigned upstream channel to another, it sends a channel change request message to the CM. In response, the CM transmits a channel change response message on the currently assigned channel to signal its readiness

to use the new channel. After switching to the new channel, the CM typically performs recalibration of transmitter parameters such as ranging offset, power level, frequency and pre-equalizer coefficients before the CM can use the new channel.

Such a channel switching mechanism can be very time-consuming and can take seconds or more because a complete re-calibration is often required.

According to this invention, a true seamless channel change can be achieved in the fsCM system 100. True seamless channel change means, on a packet-by-packet basis, each CMTS-directed cable modem burst transmission can be at any one of the upstream channels, configured with any one of the burst profiles as defined in the MAC message MDCD 1000.

The fsCM 106 joins the fsCM domain accepting new registrations in the MDCD message 1000, which also contains fields for a list of downstream channels with channel profile parameters, a list of upstream channel parameters and channel profile parameters, and a list of burst profile parameters. These profile parameters are uniquely identified within the fsMAC domain using downstream, upstream and burst ID's. These parameters are stored in the fsCM, together with the channel calibration parameters for each channel as a result of calibration request/response process.

When an upstream transmission grant is received from the MMAP 900, the grant contains sufficient information about transmission channel ID, burst profile, size of granted and header type to form an upstream burst to be transmitted at the exact start time specified in the same MMAP 900. Thus the channel change is immediate and truly seamless.

Alternative Embodiments One skill in the art can take advantage of the multi-channel fsMAC in different variations for further optimization. Examples are:

Use all downstream channels for IP packet streams, if MPEG-2 video is not needed, to further boost downstream capacity for additional users, or can be used for IP media streaming.

Use a single upstream control channel for channel calibration and bandwidth requests.

Define different upstream payload channels, such as CBR channels, dedicated channels to achieve quality of service and capacity goals.

Although the teachings of the invention have been illustrated herein in terms of a few preferred and alternative embodiments, those skilled in the art will appreciate numerous modifications, improvements and substitutions that will serve the same functions without departing from the true spirit and scope of the appended claims. All such modifications, improvement and substitutions are intended to be included within the scope of the claims appended hereto.